|Publication number||US6963871 B1|
|Application number||US 09/275,766|
|Publication date||Nov 8, 2005|
|Filing date||Mar 25, 1999|
|Priority date||Mar 25, 1998|
|Also published as||US8041560, US20050273468, US20070005567, US20080312909|
|Publication number||09275766, 275766, US 6963871 B1, US 6963871B1, US-B1-6963871, US6963871 B1, US6963871B1|
|Inventors||John Christian Hermansen, Leonard Arthur Shaefer, JR., Heather McCallum-Bayliss, Richard D. Lutz|
|Original Assignee||Language Analysis Systems, Inc.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (26), Non-Patent Citations (8), Referenced by (130), Classifications (18), Legal Events (8)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This application claims the benefit of U.S. Provisional Patent Application Ser. No. 60/079,233, filed Mar. 25, 1998, the entire disclosure of which is incorporated herein by reference.
A portion of this disclosure contains material in which copyright is claimed by the applicant and/or others. The copyright owner has no objection to the copying of this material in the course of making copies of the application file or any patents that may issue on the application, but all other rights whatsoever in the copyrighted material are reserved.
The present invention relates generally to automatic data processing systems that search and retrieve records from a database based on matching of personal names, and to improved systems and methods for intelligently processing name comparisons.
Information about individuals is often stored in a computer. Access to that information is most readily gained by using the name of the individual involved. The nature of names, however, their behavior and permutations, pose significant challenges to information retrieval. Names vary during one's life (e.g., through marriage or professional preparation); they take on different forms, depending on the formality of the situation (WILLIAM CARVER/BILLY CARVER); they may be spelled differently if recorded by someone other than the individual (PRICE/PRIES). To amplify the difficulties even more, naming conventions vary across cultures. It may not be appropriate to assume that the typical American name structure of single given name (first name), single middle name or initial followed by a surname (last name) applies in a database that contains names from all over the world, a situation that is usual in today's world of global technology and communication. Names from other cultures may have compound surnames or may be composed of only one name. Names written in writing systems other than Roman may be transcribed in a variety ways into the Roman alphabet because there is no single way to represent sounds that occur in another language but do not occur in English, causing significant differences in the spelling (KIM/GHIM).
Adequate information retrieval that is based on the name must anticipate the range and kinds of variation that can occur in names, both generally and in specific cultures. Other name search or information retrieval systems are generally unable to recognize or address the full range of variation in names. Some systems assume that names are static and search only for an exact match on the name. These systems cannot accommodate even the slightest spelling variations, initials or abbreviations (JOS. Z. BROWN/JOSEPH ZACHARY BROWNE). Other systems may use techniques or keys (such as Soundex or Soundex-like keys) that permit some minor spelling differences between names (DORSHER/DOERSHER) but these techniques generally fail to cope with significant variation (DOERSHER/DOESHER) or problems posed by names from non-Anglo cultures (ABDEL RAHMAN/ABDURRAMN). If cultural differences are recognized, it is typically through use of equivalency lists or tables. Some of the more common variants can be accommodated in this way, but retrieval is then limited to those items on the list and cannot accommodate new representations or random variation or keying errors (GOMEZ/BOMEZ).
For a system to reach a level of adequacy for automatic name searching, it must therefore address a diverse set of issues related to name variation. Although spelling variations can often be addressed through character-matching techniques (e.g., SMITH/SMYTH), false-positive matches can result from traditional string or character comparisons when common morphological endings, such as OVICH, occur at the end of otherwise dissimilar names (e.g., ZELENOVICH/JOVANOVICH). Transcription from foreign writing systems to the Roman writing system poses additional spelling concerns. Different character sets, dialectal variations and sounds that are not represented in Roman alphabetic form at all contribute to the possibility of multiple, and often inconsistent, representations of the same name. A single Chinese character (ideogram) can be transcribed to produce numerous roman forms that have little or no resemblance to one another due to dialectal variations. For example, the character CHANG, JANG and ZHANG are different roman representations of the exact same Chinese name, as are the names WU, MHO and ENG. Similarly, a single Arabic name can result in transcriptions as diverse as KHADHAFI, CODOFFI, QATHAFI.
Character-based systems may also be confronted with significant retrieval problems caused by names with the same pronunciation but with divergent spellings. WOOSTER, WORCHESTER, and WUSTER may all share at least one identical pronunciation and yet show very different spellings. When name data are shared orally, the speaker's pronunciation, the listener's hearing (or mishearing) of the name and the speaker's expectations about the spelling of the name will impact the final written representation of a name. For example, a telephone reservationist may record a caller's name with a variety of phonetically correct spellings, which may not correspond (and may therefore not be matched to) an existing database record for that caller.
Another common cause of name variation, which creates retrieval difficulty for name search systems, is the inclusion or exclusion of name data. Depending on the data source, names may be formal such as THOMAS EDWARD WINTHROP III, or informal such as TOM WINTHROP. An ideal name search system would be capable of correlating these two names, even though only a portion of the full name is available. To predict the relationship among variant formats of names, the system must also be able to recognize what rules govern which elements can be deleted or included or changed in different cultures. MARIA DEL CARMEN BUSTOS SAENZ will become MARLA DEL CARMEN BUSTOS DE LOPEZ, if she marries JUAN ANTONIO LOPEZ GARCIA. Predicting the relationship between these names is fundamental to retrieval success.
In many name search applications, it is important to identify variant forms of a name that are considered legitimate and to link and preserve the variations; in others, it may be appropriate to establish one form of a name and to treat all other forms as errors. Even if the data base is cleaned by linking variant forms and eliminating identifiable errors, users may search for names under yet more variations.
U.S. Pat. No. 5,040,218 to Vitale et al. discloses a voice synthesis system which attempts to identify the origin of a name to enhance pronunciation. The system first searches a dictionary for a name, and if the name is not found, uses grapheme and n-gram analysis to identify the name's likely origin. Similarly, U.S. Pat. 5,062,143 to Schmitt shows a system that identifies name origin using n-gram analysis.
U.S. Pat. 5,724,481 to Garberg et al. shows a method of matching proper names in a database using a phonemic representation.
U.S. Pat. 5,758,314 to McKenna shows an international database processing system. However, this system uses Soundex algorithms to process Unicode input for all cases, rather than providing a name searching system with culture-specific algorithms.
Design Pat. D359,480 shows an IPA-based computer keyboard, but does not disclose any use of IPA for identifying data records.
The article “Identifying Source Languages: the Case of Proper Names” by Valencia and Yvon (1997) discloses statistical models for name searching based on n-gram comparisons. The article also discloses determination of the source language and the use of different statistical models for comparisons, based on the source language.
John Hermansen, a named inventor, authored a doctoral dissertation, “Automatic Name Searching in Large Data Bases of International Names” (1985) which explores the concept of cultural differences in names. The document suggests searching using different culturally specific algorithms, but discloses only a simple n-gram based algorithm.
U.S. Pat. No. 5,485,373 to Davis et al. discloses a text searching system which relies on a Unicode representation (not a phonetic alphabet). The Davis system may vary algorithms based on the language being searched, but has no name classifier. This system is not designed to search for proper names; comparisons are performed based on a Unicode representation, which is not a phonetic alphabet.
Other patents relating generally to computerized language analysis and processing include: U.S. Pat. 5,323,316 to Kadashevich et al.; U.S. Pat. No. 5,337,232 to Sakai et al.; U.S. Pat. No. 5,369,726 to Kroeker et al.; U.S. Pat. No. 5,369,727 to Nomura et al.; U.S. Pat. No. 5,371,676 to Heemels et al.; U.S. Pat. Nos. 5,375,176 and 5,425,110 to Spitz; U.S. Pat. No. 5,377,280 to Nakayama; U.S. Pat. No. 5,432,948 to Davis et al.; U.S. Pat. No. 5,434,777 to Luciw; U.S. Pat. No. 5,440,663 to Moese et al.; U.S. Pat. No. 5,457,770 to Miyazawa; U.S. Pat. No. 5,490,061 to Tolin et al.; U.S. Pat. No. 5,515,475 to Gupta et al.; U.S. Pat. No. 5,526,463 to Gillick et al.; and U.S. Pat. No. 5,548,507 to Martino et al.
None of these earlier systems provide a satisfactory system and method for multicultural name searching. Thus, the inventors believe there is a need for an improved system and method for searching name-based records and for determining the degree of similarity between two name representations.
Therefore, it is a general object of the present invention to provide a name searching system architecture with multiple processing options, which automatically selects and uses an appropriate cultural-specific set of algorithms to search for database names and evaluate their proximity to a query name.
Another broad object of the invention is to provide a system and method implementing multi-algorithm name searching strategies, where search processing differs based on one or more of: culture, ethnicity, distribution, and language.
Another more specific object of the invention is to provide an improved system and method for conducting searches using a combination of n-gram distribution and positional properties to identify matches.
A further object of the invention is to provide an improved cultural name classifier which leads to application of an appropriate set of namne-regularizing linguistic rules that generate a standardized name based on stored cultural intelligence.
Yet another object of the invention is to provide an improved name classifier, incorporating a multi-step process, including preemptive lists, linguistic rules, n-gram analysis, and additional algorithms.
A further object of the invention is to provide an improved name searching system, incorporating segment-level pre-processing. Segmentation rules and syllabic stress rules contribute to a determination of where “white space” should appear in the name. Algorithms determine which graphemes are mapped to which phonemes (based on phonological, historical, and morphological principles).
Another significant object of the invention is to provide a name searching system and method incorporating an innovative key-searching system based on the International Phonetic Alphabet (IPA). This technique converts the query name to a plurality of IPA representations, which are then used to select matching keys in a first pass through the database.
A further object of the invention is to provide a name searching system and method that selectively uses sets of generic and language-specific spelling rules to infer possible phonological manifestations for personal names. A unique aspect of the comparison algorithm derives a scored match based on atomic phonological features.
Additional objects and advantages of the invention will be apparent upon review of the specification, including its drawings and appendices.
The present invention provides an improved automatic data processing system for searching names and an improved process for effectively searching and retrieving personal names in a database. It also provides a mechanism for a user to determine the distance between two names, i.e., how closely two personal names match.
In one aspect of the invention, “fuzzy logic” name searching and matching technology is provided to locate a target database record despite a lack of absolute identity between a query name and a record name.
In one preferred embodiment, a complete automated name searching system is provided, incorporating various advantageous features of the present invention. The automated search system incorporates an automatic name classifier, a multi-path architecture in which different algorithms are applied based on cultural identity of the name, name variant generation, query regularization and expansion, compensation for transpositions, affixes, and inversions, and sorting and filtering of output. The name classifier incorporates a preemptive list, analysis of morphological elements, length, and linguistic rules. The name regularizer produces a computer recognized form (character based computational representation) rather than a human recognizable form of the name. The software design uses a pronunciation equivalent (e.g. IPA) representation and language specific rules to generate name searching keys, which are used in a first pass to eliminate database entries which are obviously not matches for the name of interest.
In another preferred embodiment, the inventive search methodologies are implemented as Application Program Interfaces (APIs) that can be integrated into an existing program application or can be used to provide the foundation for a new program application that requires name matching capabilities. In API form, the features of the present invention may be selectively used in various combinations depending on the requirements of the particular application. A callable set of library routines include an intelligent preprocessor and a name evaluator that produces a score comparing a query name and database name, based on a variety of user-adjustable parameters. The user-controlled parameters permit tuning of the search methodologies for specific custom applications, so as to achieve desired levels of precision and recall for name searching in widely varying operational settings.
Referring now to
Processing of a query begins with evaluation of the searched name by name classifier module 102. Name classifier module 102 evaluates spelling, word segmentation, titles, prefixes and suffixes, and other identifiable features of the name to determine whether it falls into one of a predetermined set of identified cultural origins, including, for example, Chinese, Arabic, Hispanic, or Russian. Anglo names and names which do not fall into one of the predetermined set of special-case cultures are classified as “other” and processed according to a generic cultural algorithm.
Appropriate pre-processing is also performed to segment the name appropriately (standardize the handling of spacing between name segments and the order of the segments) and identify apparent surnames, given names, honorifics, etc., that are part in of the input name. The operation of the name classifier in this regard is unique and inventive. To determine the type of name, name classifier module 102 may use one or more of the following, depending on the observed characteristics of the name in question: a list of names which occur with high frequency in various cultures (used to preemptively type common names without extensive algorithmic processing), culturespecific linguistic rules in the form of a Titles, Affixes, and Qualifiers (TAQ) lookup table, n-gram based name typing, and name length. N-gram name typing according to the present invention may be performed as a digraph, trigraph or other n-gram analysis where both positional and distributional properties of the n-grams (e.g., digraphs and trigraphs) are used in the algorithm for making the type determination. Name classifier module 102 preferably operates according to the software design description in Appendix A, which forms a part of this specification.
LID processor 304 incorporates linguistic information aggregator 308 and LID intermediate decision processor 310. LID aggregator 308 includes high frequency name processor 316, morphological processor 318, title/affix/qualifier (TAQ) processor 320, and ngram processor 322.
LID processor 304 accumulates and weighs factors from multiple knowledge sources to determine whether there is sufficient evidence to identify the input name as belonging to a particular ethnicity, e.g. Hispanic, Arabic, etc. Linguistic information aggregator 308 performs linguistic analysis, gathering information and scoring for the input name. In the preferred embodiment, linguistic information aggregator 308 generates scores from four data sources. High frequency name processor 316 accesses a high frequency name data store of names that occur frequently in particular cultures. A match with one of these names causes aggregator 308 to retrieve and record the culture associated with the name and a confidence score associated with that name. TAQ processor 320 breaks the name into particles and makes use of the information contained in those particles to match a list of titles, affixes, and qualifiers commonly used in names of various cultures, to help determine cultural affinity. The input name is segmented based on spaces in the name, and for each segment present in the input name, TAQ processor 320 determines whether that segment is a particle present in a TAQ data store. If so, TAQ processor 320 retrieves and records the culture, name field, and confidence score associated with that TAQ particle.
Morphological processor 318 processes morphological elements such as “-ovich” which suggest a particular cultural affinity. Morphological processor 318 determines whether morphemes in a morpheme data store are present in the input name by searching for matching substrings of the name segments in the input name. For each morpheme found in the input name, morphological processor 318 records the morpheme found, the culture, name field, and confidence level associated with that morpheme.
N-gram processor 322 searches the input name for strings of letters that occur with statistical significance in names with a given cultural affinity. For each n-gram present in an associated n-gram data store, n-gram processor 322 determines whether that n-gram is present in the input name. When a match is found, the processor records the n-gram found, the culture, name field, and score associated with that n-gram.
To avoid conflict between treatment of name segments and particles by the various processing modules operating on the input name, an order of precedence is established for processing. The order of precedence is preferably TAQ particle, morpheme, and then n-gram. That is, if a string of letters is identified as aTAQ particle, that string or any substring cannot also be identified as a morpheme or n-gram for that culture. If a string is identified as a morpheme, that string and its substrings cannot be considered as part of an n-gram for that culture. Locating the name among the high frequency names for a culture does not preclude morpheme or n-grams processing of the high frequency name, but if the confidence level in the high frequency match is high, further processing may not be necessary.
As shown in
LID intermediate decision processor 310 makes a preliminary decision about the cultural affinity of the name, based on the scoring information gathered by linguistic information aggregator 308. Processor 310 determines whether enough linguistic information has been gathered by LIA 308 to confidently determine that the input name belongs to one of the cultures identified by the system. Processor 310 accepts as input one aggregate LID score for each culture, as well as an aggregate LID score for “other.” For each score, processor 310 compares the score to a LID threshold for the appropriate culture. If the LID score for a culture exceeds the threshold for that culture, processor 310 returns a value of “true” for the indicated culture. A “true” value for a culture indicates that enough evidence has been gathered to confidently identify the name as belonging to that culture. A “false” value for a culture indicates that not enough evidence has been accumulated to suggest that the name belongs to that culture. Alternatively, processor 310 may return a value for each culture equal to the LID score minus the LID threshold for that culture; in this case, negative values correspond to “false” and positive values correspond to a “true” indication.
Names which are strongly associated with one culture based on the output of LID intermediate decision processor 310 will not be processed further to identify their cultural origin, i.e. digraph and other analysis will be skipped.
Assuming the name has not been definitely identified, the surname portion is processed by digraph distribution processor 306. Based on a statistical model derived from digraph distribution statistics for names within various cultures, processor 306 computes a likelihood that the input name has a particular cultural origin. The information gathered from LID and digraph processing is combined, along with any other available information on the person (such as country of birth), in final decision processor 309. The available factors are weighted according to their confidence level to maximize the likelihood of an accurate ethnic origin evaluation. The result is an output indicating the likely classification of the name.
Following name typing, the system executes name variant generation module 104, which pre-processes the names according to culture-specific rules to generate query regularizations, based on algorithms adapted specifically for the cultural origin of the name in question, as determined by the name classifier. Variant generation module 104 also generates query expansions, i.e., identifies expected variants of the name to enhance matching capability.
As noted above, preferably, specialized processing is provided for each of a variety of ethnic name origins. Appendices B and C, which form a part of this specification, are software design descriptions for preprocessing and search algorithms for Arabic and Hispanic type names, respectively. As an example of such processing, the Hispanic processing algorithm referenced in Appendix C will now be discussed in some detail.
The process begins with a name length determiner operation on Block 702, which determines the length of the surname. Next, the name is processed by a Hispanic surname segmenter in block 704. This operation divides surnames exceeding a predetermined length (e.g. nine characters) into component segments to compensate for the fact that fixed size data fields often do not accommodate an entire Hispanic surname, leading data entry operators to conjoin name segments in a single field. Then, additional query records are generated for the separated segments and alias records are added for the separated surname segments. This process accesses a high frequency surname type data store to identify surname portions that should be separated. For example, this operation would separate “RAMIREZDELAPAZ” in the surname field into RAMIREZ DELA PAZ and “PEREZDELOPEZ” into PEREZ DE LOPEZ by finding the known surname components DE and DELA.
An Hispanic TAQ processor operates in Block 706 to scan the given name and surname for known titles, affixes, and qualifiers which do not have useful search value. TAQ elements such as DEL, DELA, DE, and SAN are then flagged to be either deleted, disregarded during matching operations, or removed. Delete means that the segment is disregarded for the remainder of the name search process and contributes marginal information to the filter process, but is not actually removed from the record. Disregard means the segment is disregarded in the remainder of the name search process but contributes to evaluation in the filter process. Remove means that a segment conjoined to the name stem is removed from the stem, and then flagged to be either deleted or disregarded as appropriate.
The Hispanic segment positioner in Block 708 operates to move any high frequency surname found in the given name field into the surname field. The name is then formatted by an Hispanic name formatter in Block 710 to generate additional name formats in case the record has more than two surname stems. Next, the name is processed by a segment position identifier in Block 712 to identify the relative position of each of the surname and given name stems. Hispanic names generally contain more than one stem in the given name and surname. In a given name, the leftmost name stem generally indicates gender; in a surname, the leftmost stem is the family name and the other stems are differentiators. Therefore, it is important to identify names that are out of position so that this may be corrected and their relevance appropriately evaluated during the search.
Next, the likely gender of the name is identified by a Hispanic gender identifier in Block 714. The gender identifier attempts to predict gender based on the gender marker of the leftmost given name segment, but may also rely on (or override the apparent gender) based on additional information such as a gender indicated as associated with the search name.
The name is processed by a frequency path director in Block 716 which directs a record for high frequency processing or low frequency processing depending on the presence or absence of high frequency surnames in the input name string.
The high frequency processor operation begins in Block 808 with generation of keys for the given names. Then, in block 810, records are retrieved according to a high frequency surname matrix and the given name keys. Control then passes to filter and sorter 610 (shown in FIG. 6).
Low frequency processor operation begins in Block 812 where each low frequency surname segment is examined to identify related high frequency and low frequency surnames, in blocks 814 and 816. This processing loop continues until names related to the segments have been identified. A “relationship” to a high frequency surname is determined by digraph comparison. If the number of identical digraphs exceeds a specified threshold, the surname is deemed to be a mere spelling variant of the similar high frequency surname. If the surnames all relate to known high frequency names, control passes through block 818 to block 808 in the high frequency processor. If the surnames have mixed high and low frequency relationships, control passes through block 820 to block 808. If all surnames have low frequency, control passes through block 822 to block 824. In block 824, a year of birth range is determined for the name. Records are then retrieved based on name content (same or different), position of the name segments, the year of birth range, the record gender, and possibly additional restrictions based on the given name.
Referring again to
Significantly, as part of name regularization for the purpose of generating an index key for a first pass through the database, the present invention applies the International Phonetic Alphabet to generate index keys, rather than using a Soundex or another conventional key. The IPA algorithm, according to the present invention, generates keys by segmenting (e.g. syllabifying) the name in question and converting it to IPA representation. In this manner, the system generates a key or set of keys which identify a set of pronounced equivalents, rather than generating a key by letter similarity, as in the traditional Soundex method. Significantly, the system generates multiple keys in IPA representation for most names, since most names have multiple possible pronunciations. The system determines multiple possible pronunciations of the name, where applicable, and associates an IPA key with each possible pronunciation. Then, records matching any of the IPA keys for a name are then selected for further consideration and comparison.
To program the IPA conversion, a rule set is generated that relates spelling to sounds. A different rule set is preferably generated for each ethnic origin of name, since pronunciations of apparently similar names may vary significantly based on origin. To generate a rule set, preferably a database of single name elements is obtained, such as a census list. The names in the list may are then manually tagged for their ethnic origin A variety of sources may then be used to determine possible pronunciations. These sources include native speaker knowledge and textual information. The rules are written broadly so that the most plausible pronunciations will be captured with some certainty. Rules for languages not written in roman characters will necessarily take into account transcription variations. The rules are written in a predetermined notation which can be processed effectively by the system. A typical rule format is:
As an example of the advantages of matching on IPA, consider a query on the name Lee. Converted to the IPA string [li], exact matches with numerous spelling variants are automatic, including Leigh and Li. Typical prior-art character based matches will fail to retrieve Leigh or Li, since the percentage of character overlap is minimal. Conversely, a standard index matching system such as Soundex will categorize Lee and Li identically, but will still miss Leigh, given the presence of a salient letter (g), and will retrieve a large number of names of low relevance, including Lu, Liao, Low, Louie, Lahoya, and Lehew. The IPA anaylsis process is further described in Appendix F, which forms a part of this specification.
While the IPA key generation, according to the present invention, provides a significant functional advantage in many cases, it should be noted that it may not be desirable to apply IPA processing to all classes of names. For example, the inventors have found that names of Arabic and Chinese origin are better processed using custom regularization algorithms rather than by the generalized IPA approach, since names acknowledged as similar in these cultures are often quite distinct phonologically.
Following regularization and expansion, name retrieval technology processing module 108 is applied. These algorithms facilitate more complete retrieval, by compensating for transpositions; deleting affixes, where appropriate; and compensating for inverted surnames, deleted surnames and nicknames. Each of these algorithms uses stored information defining naming conventions for a particular culture in the manner described herein.
Next, retrieval module 110 is applied to the results of the preprocessing performed by name classifier 102, variant generation module 104, and retrieval technology module 108. Retrieval module 110 retrieves records matching the keys (IPA or other culture-specific keys) generated by the operation of the first three modules. These records are then provided to precision filter and sorting module 112, which compares each record to the query name to determine a similarity/equivalence measurement defining the “distance” between the query name and the record name. Precision filter and sorting module 112 may perform segment position comparisons, character comparisons, phonological similarity comparisons, structural similarity comparisons, phono-feature-distance comparisons, and/or n-gram comparisons.
The output 114 of precision filter and sorting module 112 is then provided to the user. The output preferably consists of a rank-ordered list of records in descending order of likelihood of matching the query name.
One preferred embodiment implementing many desirable features of the system shown in
Another desirable implementation of the invention is as a set of name searching tools which may be provided as one or more Application Programming Interfaces (APIs) for use in developing custom database management and searching applications. A flowchart for one embodiment of an API embodiment is shown in FIG. 2. Further detail of the implementation of this embodiment is provided in Appendices K (software design description), L (default parameters), M (developer's documentation) and N (source code listing), each of which forms a part of this specification. Operation of elements in the embodiment of
As shown in
The output of ISDG 204 is search data 220, which is provided to data update and data access applications 222 and from there to the name comparison tools 212 as query and candidate search data 226. A names database with intelligent search data 224 is provided in association with data update and data access applications 222.
The embodiment of
The disclosed embodiment automatically applies whatever resources will adequately address the problem at hand, whether the variation is cross-cultural or arises from spelling variation, from transcription from other writing systems, from sound similarity, or from missing or additional information.
In operation, the user system supplies both a query name and a database name to the system. The system employs linguistic intelligence to separate the name into its integral components in intelligent pre-processor 205. Further linguistic intelligence is employed to compare the two names in name evaluator 214. The result of the comparison is a scored database name, scored name data 215. The scored name is passed to results manager 216, which collects and orders the names that are scored against a single query name. The final output is an ordered set of scored database names, ordered similar data 218.
The cornerstone of this embodiment is a programming library (functions and classes) that enables a developer to add fuzzy logic personal name searching to an application. For example, the developer may perform operations such as “Give me the 10 closest names to ‘James Slesinger’ from my database”, or “Give me all the names from my database that match ‘John Wong’ with a degree of confidence of 0.9” or “Tell me the degree of similarity between ‘Paul Vanesann’ and ‘P Vanlesann”’. The system incorporates and uses a variety of linguistic techniques to achieve these results, in the manner described previously with respect to a standalone name searching system.
Users can enhance the functionality of the APIs by incorporating their business rules and data into the name comparison process. This embodiment provides fine granularity when comparing names. That is, names are scored and ranked more precisely, which is important when dealing with large volumes of data. The technology incorporates numerous parameters (to customize the user's search comparison).
From the user/developer perspective, the name search system is quite simple to utilize. A typical name search requires the use of just four classes (SNQueryParms, SNQueryNameData, SNEvalNameData, and SNResultsList). In addition, it is important to note that the extra code required to integrate this name search technology is minimal.
The API name search interface is simplified by the fact that it makes no assumptions about the data and how it is stored. The user provides the API with the query name as well as the names from the database as input 201. The library routine then presents names which are likely matches, and qualifies their degree of similarity. From the perspective of the developer, the tool is straightforward and easy to integrate.
Searches via the API embodiment are configurable by adjusting any of 43 parameters (see Appendix L for defaults). Each parameter controls some aspect of how two names are evaluated when determining if they are similar. Some of the more basic parameters set thresholds for determining how close two names must be to be considered a match. Other parameters control more complex processing, such as how to handle multi-segment names. In general, only a small set of parameters need to be adjusted by the developer, because reasonable defaults exist for each one.
The API embodiment also provides pre-defined packages of parameters, each tailored to a particular culture or ethnicity. For example, Hispanic names have certain characteristics such as compound surnames (e.g., TORRES DE LA CRUZ) that can cause problems when searching for Hispanic names using conventional, Anglocentric methods. The Hispanic parameters package contains settings that address Hispanic-specific name issues. New cultural/ethnic parameter packages can be established and existing packages can be modified as desired.
The preferred embodiment uses a C++ object framework, so that users/developers can extend the existing product functionality to incorporate additional data elements in the scoring algorithm or create evaluation methods specific to their business or application needs. For example, a database might contain a Social Security number, in addition to given name and surname. Although the name search technology only compares name data, a developer can take advantage of class inheritance (a feature of C++), and easily subclass the program's SNEvalNameData and SNQueryNameData objects to include Social Security numbers or any other desired data element(s). These data elements can then be used in the methods that score evaluation names and determine which evaluation names are matches. In other words, record matching can be performed using name data in conjunction with other available data element information.
Users/Developers can also provide custom methods for determining if an evaluation name matches a query name or not. The default method compares the average of the given name score and surname score to a user/developer supplied threshold value. However, a more complex method may be desired. For example, the business rules of an application might dictate that a name cannot be considered a match unless either the surname or given name is an exact match. By overriding the default method, the developer can easily implement this logic in just a few lines of code.
The functions provided in the API embodiment will now be described in more detail. The available functions include comparing a query name with one or more candidate names to produce an ordered list of candidate names with the highest probability of representing the same named person. This functionality is referenced as the name comparison tools 212. The basic name checking tool employs multiple evaluation techniques to evaluate and score two names. The name checking tool incorporates information regarding variations in spelling, discrepancy in the number of name segments (amount of information included), exclusion of expected information, and positional information to establish a name score, which indicates the probability that the two names represent the same individual. The tool is controlled by a set of configurable parameters. The tool also manages and produces an ordered or unordered list of candidate names with the highest probability of representing the same named person, based on the developer defined criteria for establishing a set of results. Various culture specific callable modules are available as extensions to the name check tool, including a name classifier that culturally classifies name data, a name regularizer that levels variations in name data to a single representation, and a phonetic name key that represents name data based on phonetic similarity. Again, each of these tools and modules incorporates the methods and technology described above with reference to FIGS. 1 and 3-8.
The program also generates and stores intelligent search data for use in extracting relevant subsets of data from large data bases for further evaluation. These mechanisms will facilitate more efficient name searching while ensuring complete and accurate results. This functionality is referenced as the Name Extraction Tool(s). The disclosed embodiment provides users/developers with the capability to compare two names to determine the probability that they both represent the same named individual or to compare a single query name with a set of candidate names to determine which candidate names are most likely to represent the same named individual.
When a set of candidate names is evaluated, the APIs enable the user/developer to define the criteria for producing their own ordered list of results. The criteria for defining an ordered list of results include the following: the top X candidate names (i.e., the X candidate names scoring the highest probability that they represent the same named individual; e.g., the top ten candidate names); all candidate names whose name score exceeds a pre-defined name threshold (e.g., if the threshold =0, all candidate names will be returned in an ordered list); or the top X candidate names whose name score exceeds a pre-defined name threshold.
Name comparison tools 212 include a name evaluator 214, which employs multiple evaluation techniques to evaluate and score two names. Name evaluator 214 incorporates information regarding variations in spelling, inclusion of additional information, exclusion of expected information, and positional information in order to establish a name score, which indicates the probability that the two names represent the same individual. Name evaluator 214 is controlled by a set of configurable parameters. Results manager 216 uses the intermediate scoring information provided by name evaluator 214 to manage and produce an ordered list of candidate names with the highest probability of representing the same named person, based on the developer-defined criteria for establishing the results.
Name extraction tools 202 include an Intelligent Search Data Generator (ISDG) 204 which generates one or more search data values that facilitate extraction of relevant information from a data base for further comparative analysis. This tool is an important component of any search system that must search large volumes of data to locate similar name data, to the extent that it is not feasible to retrieve and evaluate every name record in a data base to determine its relevance to a query name. ISDG 204 provides a motivated method for retrieving all relevant information from a data base while reducing the amount of non-relevant information retrieved. This tool can provide significant performance improvements while also ensuring an accurate and complete name search. Various culture-specific tools are available as extensions to ISDG 204 to address specific issues such as the cultural classification of name data, performed by name classifier 206; leveling of variations in name data to a single representation, performed by name regularizer 208; and the representation of name data based on phonetic similarity, performed by phonetic key generator 210.
Thus, there has been disclosed an improved system and method, in multiple embodiments, for searching personal name databases, with maximum simplicity and ease of integration, maximum flexibility, and maximum extensibility.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US5040218 *||Jul 6, 1990||Aug 13, 1991||Digital Equipment Corporation||Name pronounciation by synthesizer|
|US5062143||Feb 23, 1990||Oct 29, 1991||Harris Corporation||Trigram-based method of language identification|
|US5212730 *||Jul 1, 1991||May 18, 1993||Texas Instruments Incorporated||Voice recognition of proper names using text-derived recognition models|
|US5323316||Feb 1, 1991||Jun 21, 1994||Wang Laboratories, Inc.||Morphological analyzer|
|US5337232||Mar 18, 1992||Aug 9, 1994||Nec Corporation||Morpheme analysis device|
|US5369726||Feb 9, 1993||Nov 29, 1994||Eliza Corporation||Speech recognition circuitry employing nonlinear processing speech element modeling and phoneme estimation|
|US5369727||Dec 11, 1991||Nov 29, 1994||Matsushita Electric Industrial Co., Ltd.||Method of speech recognition with correlation of similarities|
|US5371676||Jul 21, 1992||Dec 6, 1994||Oce-Nederland, B.V.||Apparatus and method for determining data of compound words|
|US5375176||Apr 19, 1993||Dec 20, 1994||Xerox Corporation||Method and apparatus for automatic character type classification of European script documents|
|US5377280||Apr 19, 1993||Dec 27, 1994||Xerox Corporation||Method and apparatus for automatic language determination of European script documents|
|US5425110||Apr 19, 1993||Jun 13, 1995||Xerox Corporation||Method and apparatus for automatic language determination of Asian language documents|
|US5432948||Apr 26, 1993||Jul 11, 1995||Taligent, Inc.||Object-oriented rule-based text input transliteration system|
|US5434777||Mar 18, 1994||Jul 18, 1995||Apple Computer, Inc.||Method and apparatus for processing natural language|
|US5440663||Jun 4, 1993||Aug 8, 1995||International Business Machines Corporation||Computer system for speech recognition|
|US5457770||Aug 19, 1993||Oct 10, 1995||Kabushiki Kaisha Meidensha||Speaker independent speech recognition system and method using neural network and/or DP matching technique|
|US5485373||Mar 25, 1993||Jan 16, 1996||Taligent, Inc.||Language-sensitive text searching system with modified Boyer-Moore process|
|US5490061||Sep 5, 1989||Feb 6, 1996||Toltran, Ltd.||Improved translation system utilizing a morphological stripping process to reduce words to their root configuration to produce reduction of database size|
|US5515475||Jun 24, 1993||May 7, 1996||Northern Telecom Limited||Speech recognition method using a two-pass search|
|US5526463||Apr 9, 1993||Jun 11, 1996||Dragon Systems, Inc.||System for processing a succession of utterances spoken in continuous or discrete form|
|US5548507||Mar 14, 1994||Aug 20, 1996||International Business Machines Corporation||Language identification process using coded language words|
|US5724481||Mar 30, 1995||Mar 3, 1998||Lucent Technologies Inc.||Method for automatic speech recognition of arbitrary spoken words|
|US5758314||May 21, 1996||May 26, 1998||Sybase, Inc.||Client/server database system with methods for improved soundex processing in a heterogeneous language environment|
|US5819265||Jul 12, 1996||Oct 6, 1998||International Business Machines Corporation||Processing names in a text|
|US5963940||Aug 14, 1996||Oct 5, 1999||Syracuse University||Natural language information retrieval system and method|
|US6311152||Apr 8, 1999||Oct 30, 2001||Kent Ridge Digital Labs||System for chinese tokenization and named entity recognition|
|USD359480||Dec 7, 1993||Jun 20, 1995||Top surface of a set of keys used for English and Japanese symbols based on International Phonetic Association coding in a keyboard configuration|
|1||*||French et al. (Applications of approximate word matching in information retrieval, ACM, 1997, pp. 9-15).|
|2||Hermansen, John Christian, "Automatic Name Searching in Large Data Bases of International Names," a Doctoral Dissertation, Georgetown University, University Microfilms International, 1985.|
|3||*||Marx et al. (Putting people first: specifying proper names in speech interfaces, ACM, 1994, pp. 29-37).|
|4||*||Oshika et al. ("Improved Retrieval Of Foreign Names From Large Database", 1998, IEEE, pp. 480-487).|
|5||*||Oshika et al. (improved retrieval of foreign names from large databases, IEEE, Feb. 1-5, 1988, pp. 480-487).|
|6||Roochnik, Paul, "Computer-Based Solutions to Certain Linquistic Problems Arising from the Romanization of Arabic Names," Doctoral Dissertation, Georgetown University, Washington, D.C., Feb. 19, 1993, pp. i-358.|
|7||Valencia, Yvon, et al., "Identification de l'Origine Linguistique: le Cas des Noms Propres", Journees Scientifiques et Techniques du Reseau Francil (JST'97), Avignon, Apr. 1997 (with English-language translation).|
|8||*||Zobel et al. (Phonetic string matching: lessons from information retrieval, ACM, 1996, pp. 166-172).|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US7089245 *||Aug 31, 2001||Aug 8, 2006||Bmc Software, Inc.||Service desk data transfer interface|
|US7395259 *||Jul 30, 2004||Jul 1, 2008||A9.Com, Inc.||Search engine system and associated content analysis methods for locating web pages with product offerings|
|US7430561||Mar 30, 2006||Sep 30, 2008||A9.Com, Inc.||Search engine system for locating web pages with product offerings|
|US7447996 *||Feb 28, 2008||Nov 4, 2008||International Business Machines Corporation||System for using gender analysis of names to assign avatars in instant messaging applications|
|US7496568 *||Nov 30, 2006||Feb 24, 2009||International Business Machines Corporation||Efficient multifaceted search in information retrieval systems|
|US7584188 *||Nov 22, 2006||Sep 1, 2009||Dun And Bradstreet||System and method for searching and matching data having ideogrammatic content|
|US7599921 *||Mar 2, 2007||Oct 6, 2009||International Business Machines Corporation||System and method for improved name matching using regularized name forms|
|US7647294 *||Oct 26, 2005||Jan 12, 2010||Fairbanks Jr William E||Indexing and querying engines and methods of indexing and querying|
|US7805430 *||Dec 22, 2005||Sep 28, 2010||Sap Ag||Evaluation of name prefix and suffix during a search|
|US7809563 *||Oct 11, 2006||Oct 5, 2010||Hyundai Autonet Co., Ltd.||Speech recognition based on initial sound extraction for navigation and name search|
|US7917509 *||Nov 5, 2007||Mar 29, 2011||Intel Corporation||String search scheme in a distributed architecture|
|US8015196||Jun 11, 2008||Sep 6, 2011||Geographic Services, Inc.||Geographic feature name search system|
|US8032532||May 21, 2008||Oct 4, 2011||International Business Machines Corporation||Efficient multifaceted search in information retrieval systems|
|US8041560||Aug 22, 2008||Oct 18, 2011||International Business Machines Corporation||System for adaptive multi-cultural searching and matching of personal names|
|US8046362||Apr 24, 2009||Oct 25, 2011||Lexisnexis Risk & Information Analytics Group, Inc.||Statistical record linkage calibration for reflexive and symmetric distance measures at the field and field value levels without the need for human interaction|
|US8090733||Jul 2, 2009||Jan 3, 2012||Lexisnexis Risk & Information Analytics Group, Inc.||Statistical measure and calibration of search criteria where one or both of the search criteria and database is incomplete|
|US8095567 *||Dec 29, 2003||Jan 10, 2012||Myfamily.Com, Inc.||Providing alternatives within a family tree systems and methods|
|US8135679||Apr 24, 2009||Mar 13, 2012||Lexisnexis Risk Solutions Fl Inc.||Statistical record linkage calibration for multi token fields without the need for human interaction|
|US8135680||Apr 24, 2009||Mar 13, 2012||Lexisnexis Risk Solutions Fl Inc.||Statistical record linkage calibration for reflexive, symmetric and transitive distance measures at the field and field value levels without the need for human interaction|
|US8135681||Apr 24, 2009||Mar 13, 2012||Lexisnexis Risk Solutions Fl Inc.||Automated calibration of negative field weighting without the need for human interaction|
|US8135719||Apr 24, 2009||Mar 13, 2012||Lexisnexis Risk Solutions Fl Inc.||Statistical record linkage calibration at the field and field value levels without the need for human interaction|
|US8190616||Jul 2, 2009||May 29, 2012||Lexisnexis Risk & Information Analytics Group Inc.||Statistical measure and calibration of reflexive, symmetric and transitive fuzzy search criteria where one or both of the search criteria and database is incomplete|
|US8195448 *||Sep 4, 2007||Jun 5, 2012||John Paisley Dargan||Method and apparatus for predicting text|
|US8195670||Apr 24, 2009||Jun 5, 2012||Lexisnexis Risk & Information Analytics Group Inc.||Automated detection of null field values and effectively null field values|
|US8229737||Jan 6, 2010||Jul 24, 2012||International Business Machines Corporation||Name classifier technique|
|US8250078||Apr 24, 2009||Aug 21, 2012||Lexisnexis Risk & Information Analytics Group Inc.||Statistical record linkage calibration for interdependent fields without the need for human interaction|
|US8260049||Sep 23, 2008||Sep 4, 2012||Abbyy Software Ltd.||Model-based method of document logical structure recognition in OCR systems|
|US8266168||Aug 8, 2008||Sep 11, 2012||Lexisnexis Risk & Information Analytics Group Inc.||Database systems and methods for linking records and entity representations with sufficiently high confidence|
|US8275770||Apr 24, 2009||Sep 25, 2012||Lexisnexis Risk & Information Analytics Group Inc.||Automated selection of generic blocking criteria|
|US8285725||Jul 2, 2009||Oct 9, 2012||Lexisnexis Risk & Information Analytics Group Inc.||System and method for identifying entity representations based on a search query using field match templates|
|US8306970||Jan 9, 2009||Nov 6, 2012||Drubner Jeffrey M||Method and system for uniquely identifying a person to the exclusion of all others|
|US8316047||Apr 24, 2009||Nov 20, 2012||Lexisnexis Risk Solutions Fl Inc.||Adaptive clustering of records and entity representations|
|US8321440||Mar 7, 2011||Nov 27, 2012||Intel Corporation||String search scheme in a distributed architecture|
|US8423563 *||Feb 23, 2004||Apr 16, 2013||Sybase, Inc.||System and methodology for name searches|
|US8452132||Mar 30, 2010||May 28, 2013||Abbyy Software Ltd.||Automatic file name generation in OCR systems|
|US8484168||Mar 12, 2012||Jul 9, 2013||Lexisnexis Risk & Information Analytics Group, Inc.||Statistical record linkage calibration for multi token fields without the need for human interaction|
|US8484211||Jul 2, 2009||Jul 9, 2013||Lexisnexis Risk Solutions Fl Inc.||Batch entity representation identification using field match templates|
|US8489617||Jun 5, 2012||Jul 16, 2013||Lexisnexis Risk Solutions Fl Inc.||Automated detection of null field values and effectively null field values|
|US8495076||Nov 18, 2011||Jul 23, 2013||Lexisnexis Risk Solutions Fl Inc.||Statistical measure and calibration of search criteria where one or both of the search criteria and database is incomplete|
|US8495077||Jul 11, 2012||Jul 23, 2013||Lexisnexis Risk Solutions Fl Inc.||Database systems and methods for linking records and entity representations with sufficiently high confidence|
|US8560403 *||Aug 10, 2007||Oct 15, 2013||Left Bank Ventures, Llc||System and method for demand driven collaborative procurement, logistics, and authenticity establishment of luxury commodities using virtual inventories|
|US8572052||Mar 12, 2012||Oct 29, 2013||LexisNexis Risk Solution FL Inc.||Automated calibration of negative field weighting without the need for human interaction|
|US8572070||Jul 2, 2009||Oct 29, 2013||LexisNexis Risk Solution FL Inc.||Statistical measure and calibration of internally inconsistent search criteria where one or both of the search criteria and database is incomplete|
|US8639691||Jul 2, 2009||Jan 28, 2014||Lexisnexis Risk Solutions Fl Inc.||System for and method of partitioning match templates|
|US8639705||Jul 2, 2009||Jan 28, 2014||Lexisnexis Risk Solutions Fl Inc.||Technique for recycling match weight calculations|
|US8661026||Jul 2, 2009||Feb 25, 2014||Lexisnexis Risk Solutions Fl Inc.||Entity representation identification using entity representation level information|
|US8706723 *||Jun 22, 2011||Apr 22, 2014||Jostle Corporation||Name-search system and method|
|US8719027 *||Feb 28, 2007||May 6, 2014||Microsoft Corporation||Name synthesis|
|US8768970||Dec 15, 2011||Jul 1, 2014||Ancestry.Com Operations Inc.||Providing alternatives within a family tree systems and methods|
|US8799234 *||Feb 3, 2011||Aug 5, 2014||Microsoft Corporation||Semantic entity manipulation using input-output examples|
|US8812300 *||Sep 22, 2011||Aug 19, 2014||International Business Machines Corporation||Identifying related names|
|US8819084||Mar 15, 2006||Aug 26, 2014||Bmc Software, Inc.||Service desk data transfer interface|
|US8855998 *||Sep 22, 2011||Oct 7, 2014||International Business Machines Corporation||Parsing culturally diverse names|
|US8861856||Aug 28, 2012||Oct 14, 2014||Abbyy Development Llc||Model-based methods of document logical structure recognition in OCR systems|
|US8972930||Jun 4, 2010||Mar 3, 2015||Microsoft Corporation||Generating text manipulation programs using input-output examples|
|US8996358||May 25, 2012||Mar 31, 2015||International Business Machines Corporation||Entity variant generation and normalization|
|US9015171||Dec 14, 2009||Apr 21, 2015||Lexisnexis Risk Management Inc.||Method and system for linking and delinking data records|
|US9020971||Sep 17, 2013||Apr 28, 2015||Lexisnexis Risk Solutions Fl Inc.||Populating entity fields based on hierarchy partial resolution|
|US9026514 *||Oct 13, 2006||May 5, 2015||International Business Machines Corporation||Method, apparatus and article for assigning a similarity measure to names|
|US9031979||Sep 17, 2013||May 12, 2015||Lexisnexis Risk Solutions Fl Inc.||External linking based on hierarchical level weightings|
|US9037606||Sep 17, 2013||May 19, 2015||Lexisnexis Risk Solutions Fl Inc.||Internal linking co-convergence using clustering with hierarchy|
|US9043359||Sep 17, 2013||May 26, 2015||Lexisnexis Risk Solutions Fl Inc.||Internal linking co-convergence using clustering with no hierarchy|
|US9063926||Aug 27, 2012||Jun 23, 2015||International Business Machines Corporation||Entity variant generation and normalization|
|US9189505||Aug 9, 2010||Nov 17, 2015||Lexisnexis Risk Data Management, Inc.||System of and method for entity representation splitting without the need for human interaction|
|US9229926||Dec 3, 2012||Jan 5, 2016||International Business Machines Corporation||Determining similarity of unfielded names using feature assignments|
|US9251140||Jul 11, 2013||Feb 2, 2016||International Business Machines Corporation||Determining similarity of unfielded names using feature assignments|
|US9275412||Sep 28, 2012||Mar 1, 2016||Verus Financial, Llc||Method and system for uniquely identifying a person to the exclusion of all others|
|US9384262||Mar 24, 2015||Jul 5, 2016||Lexisnexis Risk Solutions Fl Inc.||Internal linking co-convergence using clustering with hierarchy|
|US9411859||Mar 26, 2015||Aug 9, 2016||Lexisnexis Risk Solutions Fl Inc||External linking based on hierarchical level weightings|
|US9413912||Dec 12, 2012||Aug 9, 2016||Abbyy Development Llc||Scanning device having a bed cover including a pattern of repeated design elements|
|US9501505||Oct 20, 2015||Nov 22, 2016||Lexisnexis Risk Data Management, Inc.||System of and method for entity representation splitting without the need for human interaction|
|US9552335||Jun 4, 2012||Jan 24, 2017||Microsoft Technology Licensing, Llc||Expedited techniques for generating string manipulation programs|
|US9589073 *||Apr 28, 2014||Mar 7, 2017||Verint Systems Ltd.||Systems and methods for keyword spotting using adaptive management of multiple pattern matching algorithms|
|US9607105 *||Mar 30, 2011||Mar 28, 2017||Amazon Technologies, Inc.||Content searching techniques|
|US9613115||Jul 12, 2010||Apr 4, 2017||Microsoft Technology Licensing, Llc||Generating programs based on input-output examples using converter modules|
|US9679025||Apr 17, 2015||Jun 13, 2017||International Business Machines Corporation||Providing search query results based on entity variant generation and normalization|
|US9710569||Aug 25, 2014||Jul 18, 2017||Bmc Software, Inc.||Service desk data transfer interface|
|US20040002850 *||Mar 14, 2002||Jan 1, 2004||Shaefer Leonard Arthur||System and method for formulating reasonable spelling variations of a proper name|
|US20050004889 *||Jul 30, 2004||Jan 6, 2005||Bailey David R.||Search engine system and associated content analysis methods for locating web pages with product offerings|
|US20050084152 *||Feb 23, 2004||Apr 21, 2005||Sybase, Inc.||System and methodology for name searches|
|US20050119875 *||Sep 17, 2004||Jun 2, 2005||Shaefer Leonard Jr.||Identifying related names|
|US20050141391 *||Feb 16, 2005||Jun 30, 2005||Tetsuo Ueyama||Optical pickup|
|US20050147947 *||Dec 29, 2003||Jul 7, 2005||Myfamily.Com, Inc.||Genealogical investigation and documentation systems and methods|
|US20050149497 *||Dec 29, 2003||Jul 7, 2005||Myfamily.Com, Inc.||Providing alternatives within a family tree systems and methods|
|US20060031069 *||Aug 3, 2004||Feb 9, 2006||Sony Corporation||System and method for performing a grapheme-to-phoneme conversion|
|US20060089927 *||Oct 26, 2005||Apr 27, 2006||Tildy, Llc||Indexing and querying engines and methods of indexing and querying|
|US20060122903 *||Feb 25, 2006||Jun 8, 2006||Medrano Guillermo A||Cross-Cultural and Personality Support System and Method|
|US20060161581 *||Mar 15, 2006||Jul 20, 2006||Bmc Software, Inc.||Service desk data transfer interface|
|US20060179070 *||Mar 15, 2006||Aug 10, 2006||Bmc Software, Inc.||Service desk data transfer interface|
|US20070005567 *||Jan 25, 2002||Jan 4, 2007||Hermansen John C||System and method for adaptive multi-cultural searching and matching of personal names|
|US20070005586 *||Mar 30, 2005||Jan 4, 2007||Shaefer Leonard A Jr||Parsing culturally diverse names|
|US20070005597 *||Nov 18, 2005||Jan 4, 2007||Williams Charles K||Name classifier algorithm|
|US20070038652 *||Aug 15, 2005||Feb 15, 2007||Microsoft Corporation||Data driven cultural customization|
|US20070136070 *||Oct 11, 2006||Jun 14, 2007||Bong Woo Lee||Navigation system having name search function based on voice recognition, and method thereof|
|US20070150438 *||Dec 22, 2005||Jun 28, 2007||Sap Ag||Evaluation of name prefix and suffix during a search|
|US20070162445 *||Nov 22, 2006||Jul 12, 2007||Dun And Bradstreet||System and method for searching and matching data having ideogrammatic content|
|US20070174339 *||Sep 7, 2006||Jul 26, 2007||Kolo Brian A||System and method for determining personal genealogical relationships and geographical origins including a relative confidence|
|US20070218429 *||Sep 7, 2006||Sep 20, 2007||Kolo Brian A||System and method for determining personal genealogical relationships and geographical origins|
|US20070244911 *||Mar 21, 2006||Oct 18, 2007||Dick Richard S||Composite clinical data dictionary (CēDē)|
|US20080059465 *||Nov 5, 2007||Mar 6, 2008||Intel Corporation||String Search Scheme in a Distributed Architecture|
|US20080091674 *||Oct 13, 2006||Apr 17, 2008||Thomas Bradley Allen||Method, apparatus and article for assigning a similarity measure to names|
|US20080133473 *||Nov 30, 2006||Jun 5, 2008||Broder Andrei Z||Efficient multifaceted search in information retrieval systems|
|US20080162113 *||Sep 4, 2007||Jul 3, 2008||Dargan John P||Method and Apparatus for for Predicting Text|
|US20080208574 *||Feb 28, 2007||Aug 28, 2008||Microsoft Corporation||Name synthesis|
|US20080215562 *||Mar 2, 2007||Sep 4, 2008||David Edward Biesenbach||System and Method for Improved Name Matching Using Regularized Name Forms|
|US20080222003 *||Aug 10, 2007||Sep 11, 2008||Lottabase, Llc||System and method for demand driven collaborative procurement, logistics, and authenticity establishment of luxury commodities using virtual inventories|
|US20080222117 *||May 21, 2008||Sep 11, 2008||Broder Andrei Z||Efficient multifaceted search in information retrieval systems|
|US20080312909 *||Aug 22, 2008||Dec 18, 2008||International Business Machines Corporation||System for adaptive multi-cultural searching and matching of personal names|
|US20080319990 *||Jun 11, 2008||Dec 25, 2008||Geographic Services, Inc.||Geographic feature name search system|
|US20090087094 *||Sep 23, 2008||Apr 2, 2009||Dmitry Deryagin||Model-based method of document logical structure recognition in ocr systems|
|US20090222255 *||Jul 31, 2008||Sep 3, 2009||International Business Machines Corporation||Using gender analysis of names to assign avatars in instant messaging applications|
|US20090222526 *||Jul 31, 2008||Sep 3, 2009||International Business Machines Corporation||Using gender analysis of names to assign avatars in instant messaging applications|
|US20100114812 *||Jan 6, 2010||May 6, 2010||International Business Machines Corporation||Name classifier technique|
|US20100211380 *||Jan 15, 2010||Aug 19, 2010||Sony Corporation||Information processing apparatus and information processing method, and program|
|US20100215272 *||Mar 30, 2010||Aug 26, 2010||Andrey Isaev||Automatic file name generation in ocr systems|
|US20100281020 *||Jan 9, 2009||Nov 4, 2010||Drubner Jeffrey M||Method and system for uniquely identifying a person to the exclusion of all others|
|US20110173232 *||Mar 7, 2011||Jul 14, 2011||Intel Corporation||String search scheme in a distributed architecture|
|US20120011084 *||Feb 3, 2011||Jan 12, 2012||Microsoft Corporation||Semantic entity manipulation using input-output examples|
|US20120016660 *||Sep 22, 2011||Jan 19, 2012||International Business Machines Corporation||Parsing culturally diverse names|
|US20120016663 *||Sep 22, 2011||Jan 19, 2012||International Business Machines Corporation||Identifying related names|
|US20120330947 *||Jun 22, 2011||Dec 27, 2012||Jostle Corporation||Name-Search System and Method|
|US20130268564 *||Jan 11, 2013||Oct 10, 2013||Ancestry.Com Operations Inc.||Genealogy investigation and documentation systems and methods|
|US20150255069 *||Mar 4, 2014||Sep 10, 2015||Amazon Technologies, Inc.||Predicting pronunciation in speech recognition|
|US20150310014 *||Apr 28, 2014||Oct 29, 2015||Verint Systems Ltd.||Systems and methods for keyword spotting using adaptive management of multiple pattern matching algorithms|
|CN100489853C||Dec 28, 2006||May 20, 2009||腾讯科技（深圳）有限公司||System and method for quickly inquiring about black and white name list|
|CN102223430A *||Jun 13, 2011||Oct 19, 2011||深圳桑菲消费通信有限公司||Method for ranking and searching polyphones of contacts in mobile phone|
|CN103472991A *||Aug 28, 2013||Dec 25, 2013||华为技术有限公司||Method, equipment and system for adding contacts|
|EP1808781A2 *||Nov 30, 2006||Jul 18, 2007||Sap Ag||Evaluation of name prefix and suffix during a search|
|EP1808781A3 *||Nov 30, 2006||Aug 22, 2007||Sap Ag||Evaluation of name prefix and suffix during a search|
|WO2010003061A1 *||Jul 2, 2009||Jan 7, 2010||Lexis Nexis Risk & Information Analytics Group, Inc.||Database systems and methods|
|U.S. Classification||1/1, 707/E17.039, 707/E17.061, 704/1, 707/999.006, 707/999.003, 707/999.101, 707/999.104|
|International Classification||G06F7/00, G06F17/30|
|Cooperative Classification||Y10S707/99933, Y10S707/99945, Y10S707/99942, Y10S707/99936, G06F17/30634, G06F17/30985|
|European Classification||G06F17/30Z2P5, G06F17/30T2|
|Mar 17, 2000||AS||Assignment|
Owner name: LANGUAGE ANALYSIS SYSTEMS, INC., VIRGINIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:HERMANSEN, JOHN CHRISTIAN;SHAEFER, LEONARD ARTHUR JR.;MCCALLUM-BAYLISS, HEATHER;AND OTHERS;REEL/FRAME:010620/0315
Effective date: 20000317
|Aug 23, 2002||AS||Assignment|
Owner name: LANGUAGE ANALYSIS SYSTEMS FEDERAL CONSULTING, INC.
Free format text: CHANGE OF NAME;ASSIGNOR:LANGUAGE ANALYSIS SYSTEMS, INC.;REEL/FRAME:013225/0867
Effective date: 20020701
|Apr 30, 2004||AS||Assignment|
Owner name: LANGUAGE ANALYSIS SYSTEMS, INC., VIRGINIA
Free format text: CHANGE OF NAME;ASSIGNOR:LANGUAGE ANALYSIS SYSTEMS FEDERAL CONSULTING, INC.;REEL/FRAME:015281/0712
Effective date: 20020806
|Aug 5, 2004||AS||Assignment|
Owner name: LANGUAGE ANALYSIS SYSTEMS, INC., VIRGINIA
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:LANGUAGE ANALYSIS SYSTEMS FEDERAL CONSULTING, INC.;REEL/FRAME:015650/0167
Effective date: 20020806
|Nov 17, 2006||AS||Assignment|
Owner name: IBM CORPORATION, NEW YORK
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNOR:LANGUAGE ANALYSIS SYSTEMS, INC.;REEL/FRAME:018532/0089
Effective date: 20060821
|May 18, 2009||REMI||Maintenance fee reminder mailed|
|Nov 8, 2009||LAPS||Lapse for failure to pay maintenance fees|
|Dec 29, 2009||FP||Expired due to failure to pay maintenance fee|
Effective date: 20091108